It doesn't need to do it between billion of pages. As soon as they detect duplicates, there is a high chance that it will be done again. So monitoring the two or more site with same content is doable but not very efficient. Ideally, the author subject to scrapping should be able to reference its page to google to prove anteriority. The scraper won't be able to do that.